Zishu Yao

Zishu Yao

Auto-generated profile, based on publicly available information

Zishu Yao

College of Computer and Data Science, Fuzhou University, Fuzhou, China

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
29 July 2024 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
ZYZishu Yao
Zishu Yao
GFGuodong Fan
Guodong Fan
JFJinfu Fan
Jinfu Fan
MGMin Gan
Min Gan
CCC.L. Philip Chen
C.L. Philip Chen
Low-light remote sensing images generally feature high resolution and high spatial complexity, with continuously distributed surface features in space. This continuity in scenes leads to extensive long-range correlations in spatial domains within remote sensing images. Convolutional Neural Networks, which rely on local correlations for long-distance modeling, struggle to establish long-range correlations in such images. On the other hand, transformer-based methods that focus on global information face high computational complexities when processing high-resolution remote sensing images. From another perspective, Fourier transform can compute global information without introducing a large number of parameters, enabling the network to more efficiently capture the overall image structure and establish long-range correlations. Therefore, we propose a Dual-Domain Feature Fusion Network (DFFN) for low-light remote sensing image enhancement. Specifically, this challenging task of low-light enhancement is divided into two more manageable sub-tasks: the first phase learns amplitude information to restore image brightness, and the second phase learns phase information to refine details. To facilitate information exchange between the two phases, we designed an information fusion affine block that combines data from different phases and scales. Additionally, we have constructed two dark light remote sensing datasets to address the current lack of datasets in dark light remote sensing image enhancement. Extensive evaluations show that our method outperforms existing state-of-the-art methods. The code is available at https://github.com/iijjlk/DFFN.

12 January 2024 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
ZYZishu Yao
Zishu Yao
JSJian-Nan Su
Jian-Nan Su
GFGuodong Fan
Guodong Fan
MGMin Gan
Min Gan
CCC.L. Philip Chen
C.L. Philip Chen
Image gradients contain crucial information in the images. However, the gradient information of low-light images is often concealed in darkness and is susceptible to noise contamination. This imprecise gradient information poses a significant obstacle to low-light image enhancement (LLIE) tasks. Simultaneously, methods relying solely on pixel-level reconstruction loss struggle to accurately correct the mapping from dimly lit images to normal images, resulting in restored outcomes with color abnormalities or artifacts. In this article, we propose a gradient-aware and contrastive-adaptive (GACA) learning framework to address the aforementioned issues. GACA initially estimates more accurate gradient information and employs it as a structural prior to guide image generation. Simultaneously, we introduce a novel regularization constraint to better rectify the image mapping. Extensive experiments on benchmark datasets and downstream segmentation tasks demonstrate the state-of-the-art performance and generalization. Compared to existing approaches, our method achieves an average 4.7% reduction in natural image quality evaluator (NIQE) on benchmark datasets. The code is available at https://github.com/iijjlk/GACA .

1 January 2025 in Computer Vision and Image Understanding
Verified
Journal Article
Closed access
GFGuodong Fan
Guodong Fan
ZYZishu Yao
Zishu Yao
MGMin Gan
Min Gan

17 September 2025 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
BZBosong Zhuang
Bosong Zhuang
ZYZishu Yao
Zishu Yao
GFGuodong Fan
Guodong Fan
JLJinjiang Li
Jinjiang Li
Low-light remote sensing images typically cover vast areas. They contain objects of various scales and have localized light sources. This makes it challenging to enhance brightness while preserving fine image structures. Existing approaches are primarily designed in the spatial domain. However, due to the tight coupling between illumination degradation and structural information, these methods often struggle to achieve effective enhancement. In this paper, we propose a divide-and-conquer frequency domain decoupling enhancement strategy. Specifically, by exploring the decoupling properties of the frequency domain, we design a Light Contrastive Regularization that constrains the model to focus solely on brightness distribution in the contrastive space while reducing interference from redundant information. Additionally, we introduce a novel Phase Mamba Enhancement Network, which leverages the unique continuity of the frequency domain. By employing a continuous scanning mechanism, our model effectively captures long-range dependencies in low-light remote sensing images, enabling finer-grained structural restoration. Extensive experiments demonstrate that our method surpasses state-of-the-art approaches both qualitatively and quantitatively.

18 February 2026 in Pattern Recognition
Verified
Journal Article
Closed access
ZYZishu Yao
Zishu Yao
XSXiang-Xiang Su
Xiang-Xiang Su
SZShengning Zhou
Shengning Zhou
GZGuangyu Zhu
Guangyu Zhu
GFGuodong Fan
Guodong Fan
JCJing Chen
Jing Chen

13 April 2026 in IEEE Transactions on Circuits and Systems for Video Technology
Verified
Journal Article
Closed access
RSRui-Lin Shi
Rui-Lin Shi
ZYZi-Shu Yao
Zi-Shu Yao
FCFei-Yan Chen
Fei-Yan Chen
GCGuang-Yong Chen
Guang-Yong Chen
MGMin Gan
Min Gan
CCC.L. Philip Chen
C.L. Philip Chen
Scene Text Image Super-Resolution (STISR) methods improve recognition accuracy by refining text regions locally. However, most existing approaches are limited by short-range dependencies, hindering the modeling of long-range semantic relationships between characters, which reduces their effectiveness in complex text scenes. Moreover, current methods are predominantly evaluated on synthetic datasets, which do not adequately capture real-world challenges such as diverse text styles, complex backgrounds, and spatial distortions. To address these limitations, we propose a Dual-Guided Visual and Semantic (DGVS) framework. This innovative framework utilizes a recognizer-driven attention mechanism to decouple character sequences, effectively distinguishing text from background noise. Additionally, we incorporate a state-space model to establish global semantic reasoning links, enhancing the comprehension of contextual relationships within text images. Furthermore, we construct Real-World Text (RealWT), a novel real-world benchmark dataset that integrates diverse data sources, including online images and multi-device captures. This dataset incorporates factors like device variations, resolution differences, and degradation, offering a more realistic simulation of real-world conditions. It enables models to learn degradation patterns that closely mirror practical applications, offering a standardized benchmark for evaluating STISR performance. Extensive experiments demonstrate that our method outperforms existing approaches, as validated by evaluations on TextZoom and RealWT. Our dataset and code are available on https://github.com/yoursmith/sde-DGVS.